About
Work
Discera
Noesira
Links
The motto
Every task has a different winner.
What my research company, Noesira, is built on.
GPU NODE 01
U01
U02
U03
U04
U05
CABLE MGMT
GPU NODE 02
U01
U02
U03
U04
U05
CABLE MGMT
GPU NODE 03
U01
U02
U03
U04
U05
CABLE MGMT
42U RACK
BENCH 03
COLD AISLE C-3
GPU
PCB REV 2.1
M1
M2
M3
VRM
8-PIN
PWM
92MM
SN 0424
L2
L2
RF
L1$
SCH
SM0
RF
L1$
SCH
SM1
RF
L1$
SCH
SM2
RF
L1$
SCH
SM3
RF
L1$
SCH
SM4
RF
L1$
SCH
SM5
RF
L1$
SCH
SM6
RF
L1$
SCH
SM7
RF
L1$
SCH
SM8
PLL
0.31 -1.07 0.88
0.12 -0.58 1.84
0.73
W
OUT
reply
AI DATA CENTER
HV
130 FT
FACILITY 07 · ILLUSTRATED
Inside an AI · illustrated tour
01 / 06
Noesira field trip · illustrated tour
Inside an AI
An illustrated journey from computing hardware to a model’s reply, in six stops. Keep scrolling.
Ask one of the big AI services a question, and the work starts in a building like this: a data center packed with computers running day and night. Inside: rack after rack of GPUs — chips built to do the same kind of math over and over, at enormous scale. One of those GPUs: it spreads the work across thousands of small cores, all computing at the same time. Deeper, on the chip itself: the model — a vast web of connections, each one adjusted during training on enormous amounts of data. Every connection carries a number the model learned — a weight. Your prompt is turned into tokens and processed through billions of them. The numbers converge on a reply, sent back one token at a time. Noesira’s question: which model does it best?
01 About short version
I'm Jack, and I build AI products.
I run Discera , a router that looks at each message and sends it to the AI that fits the job best.
Noesira is my independent AI research company. It ranks AI models by what they're best at, from public benchmarks, with sources published.
Writing Reasoning Code Images Live Search Long Context Routing Writing Reasoning Code Images Live Search Long Context Routing
03 Discera the AI that chooses for you
I built Discera because no single model is best at everything. It's one prompt box that reads your task and sends it to the model that handles it best.
?
WRITING
CODE
IMAGES
One prompt in, the right model out
Fig. 02
Discera reports 94.1% average routing accuracy across three 147-question test sets, versus 73.7% for the original router it was measured against on the same benchmark. As published on discera.chat
Visit Discera discera.chat (access is limited by code) ↗
04 Noesira independent AI research
I started Noesira to rank AI models by what they're actually best at: writing, coding, reasoning, research, routing. It's compiled from public benchmarks and arenas, and published with the sources attached.
Visit Noesira noesira.org ↗
05 Links find the companies
Jack Bailey